Short answer: a dermatology clinic's chat widget has to do two jobs at once — route medical skin enquiries to a human quickly and correctly, and handle the high-volume cosmetic questions about price, downtime and process itself. The hard rule is unchanged either way: no assessment of anyone's skin, ever, and no response to photos. Everything clinical becomes a captured enquiry for a dermatologist's team.
Two waiting rooms, one website
Most dermatology practices run a medical side and a cosmetic side, and the website is where they collide. A visitor might be worried about a changing mole, or they might be pricing up a laser resurfacing course. Same site, same chat window, completely different handling.
The medical stream needs speed and a clean handoff. Someone concerned about a lesion, a persistent rash, a suspected skin cancer or a flare of an existing condition should get one thing from the assistant: a clear statement that a dermatologist needs to see this, how to get an appointment, what to bring, whether a referral is required, and — critically — an immediate capture of their details so the clinic can act. The assistant should not ask what it looks like, how long it has been there, whether it has changed, or anything else that resembles history taking.
The cosmetic stream is where an AI chatbot does most of its visible work, because those questions are logistics dressed as clinical questions and they arrive in volume.
The cosmetic question set
What does a treatment cost, and per session or per course. How many sessions are typical. How long is the appointment. Is there downtime, and how much. Is it painful, do you numb. When do results show. How long do they last. Do I need a patch test. Do I need a consultation first, and is it charged. Who performs it — consultant, doctor, nurse. What are the aftercare requirements. Can I have it in summer. What is your cancellation policy.
Every one of those is a published fact about how your clinic operates. Clerkzo crawls your site to learn them, and the ones that live only in staff heads — your patch test lead time, which practitioner does which laser, your policy on treating tanned skin — go in as knowledge file uploads.
What the assistant must not do is bridge from those facts to a person. It cannot say whether a treatment suits someone's skin type, whether their condition would respond, whether a mole is concerning, or what a photo shows. Photos in particular: the correct behaviour is to decline to interpret and offer a consultation. Clerkzo's guardrails and human handoff make that the built-in path rather than something you rely on the model to choose.
Referrals, insurance and the paperwork wall
Dermatology has an admin layer that quietly absorbs staff hours: do you need a GP referral, do you take self-pay, which insurers do you deal with, do you need pre-authorisation, what is the code for this procedure, can you write to my insurer, how long is the wait for a medical appointment versus a cosmetic one, do you do teledermatology.
None of that is clinical and all of it is answerable. Publishing it clearly and letting the assistant state it removes a large share of the phone traffic and, more usefully, removes the enquiries that were never going to convert because the clinic does not take that insurer or does not offer that route.
The same pattern shows up across med spas and other clinics that mix funded and self-pay work: publish the clinic-level facts, and let a person handle anything specific to one policy.
Capturing the request
Whichever stream a conversation belongs to, it should end with a captured request rather than a suggestion to call.
Clerkzo collects leads conversationally or as an in-chat form with custom fields — name, phone, email, medical or cosmetic, treatment or concern in general terms, existing patient or new, insurance or self-pay, preferred times. Keep the concern field short and non-clinical; the point is routing, not history.
Requests land in a leads inbox with a New to Contacted to Qualified to Won or Lost pipeline. Set up notifications so medical enquiries reach the right person fast — email, webhook or Telegram, and webhook if you want them pushed into your own system automatically.
Clerkzo does not write to your calendar. It captures the booking request and hands it to your team, which is essential in a clinic where a fifteen minute cosmetic review and a forty minute mole mapping appointment cannot be treated the same way. Your scheduling software stays authoritative. See appointment booking.
What changes in the clinic
The most common report from clinics that do this well is not more enquiries. It is cleaner ones. Cosmetic enquiries arrive already knowing the price and the downtime, so consultations convert better. Medical enquiries arrive with contact details attached instead of as an anonymous voicemail. And the evening flood of price questions — which is real, because cosmetic research happens at night — no longer waits until morning to get an answer. After-hours answering is the same idea generalised.
Your reception team still does the work that requires judgement. What disappears is the twentieth explanation of what a course of laser involves.
Setup and governance
One script tag, live in roughly ten minutes, trained by crawling your site, themed to match your brand with Widget Studio. Flat pricing from nineteen dollars a month with a free plan, and done-for-you setup if you want help configuring the medical and cosmetic boundaries properly.
Clerkzo is SOC 2 Type II examined, GDPR compliant and HIPAA assessed. Before launch, have a clinician read a set of test transcripts and confirm the assistant refuses to assess in every phrasing they can think of. Then use the corrections loop for anything that reads wrong.
Frequently asked questions
Can it look at a photo of a mole?
No. The assistant declines to interpret images and directs the person to a dermatology appointment, capturing their details in the process.
Can it tell someone if their condition is treatable at the clinic?
It can state which conditions and treatments the clinic offers as published services. Whether a specific person is suitable is a clinical judgement.
How do we keep urgent medical enquiries from sitting in a queue?
Route them with their own notification path — a webhook or Telegram channel your clinical admin watches — separate from cosmetic enquiries.
Does it replace our referral process?
No. It explains what your referral process requires and captures the enquiry. The process itself stays with your team.